I believe it will do both, it just depends on the measure you use. In an project there are three measures of success or failure, cost, time-scale and outcome and I believe it will fail on two but succeed on the most important and have set out my argument in a blog post here https://communities.theiet.org/groups/blogpost/view/27/231/6920
The project is so complex to think costs will not overrun or timing slip is to be naive, as it is impossible to predict them when the timescales are so long and the complexity so great, but the outcome will be a success
